Approximation of Optimized Fuzzy Logic Controller for Shunt Active Power Filter

Abstract

The exponential growth of power electronic controlled equipment and non-linear loads have given rise to a type of voltage and current waveform distortion, termed as ‘harmonics’, adversely affecting the power quality (PQ). Moreover, the sensitivity of these equipments to PQ disturbances has motivated the researchers to develop dynamic and adjustable solutions for harmonic mitigation. Active power filters (APFs) address almost each attribute of PQ depending upon the topology used. The current controlled voltage source inverter (VSI) based shunt active power filter (SAPF) emerges out to be an undisputed alternative for current harmonic mitigation. SAPF having pulse width modulation (PWM) controlled voltage source inverter (VSI) topology is extensively used in distribution power systems, which conventionally utilizes the PI controller for reference voltage tracking. In recent times, Fuzzy logic controllers (FLCs) have been established as viable alternatives of conventional PI controllers in highly non-linear control applications, with varying operating conditions. The improved performance of conventionally used large rule FLC is achieved at the cost of increased complexity, leading to large computational time, and memory requirement. Conventionally triangular membership functions (MFs) are used to represent input and output variables of an FLC. In this chapter other less explored MFs such as generalized bell (Gbell), Gaussian and difference sigmoid (Dsig) are also investigated to find optimal membership function. Gaussian MFs based FLC evolves as the optimized FLC in terms of providing effective harmonic compensation along with efficient dynamic response under randomly varying loading conditions. The chapter focuses on three main areas, i.e., PQ problem of current harmonics and its mitigation, selection of optimized FLC for shunt APF and complexity reduction of optimized FLC using an approximation technique. The proposed approximation is based on minimizing the sum of square errors, between the outputs of large rule FLC and simplest 4-rule FLC. This approximation of large rules optimized FLC results in reduced computational and functional complexity and less memory requirement without compromising the control performances of FLC in terms of dynamic response and harmonic compensation capabilities. Proposed approximation technique considerably improves the harmonic compensation performance of shunt APF, due to effective approximation and smoother transition of output in the entire UOD.